Extract random 2d windows of a 2d numpy array - python-3.x

import numpy as np
arr = np.array(range(60)).reshape(6,10)
arr
> array([[ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9],
> [10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
> [20, 21, 22, 23, 24, 25, 26, 27, 28, 29],
> [30, 31, 32, 33, 34, 35, 36, 37, 38, 39],
> [40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
> [50, 51, 52, 53, 54, 55, 56, 57, 58, 59]])
What I need:
select_random_windows(arr, number_of windows= 3, window_size=3)
> array([[[ 1, 2, 3],
> [11, 12, 13],
> [21, 22, 23]],
>
> [37, 38, 39],
> [47, 48, 49],
> [57, 58, 59]],
>
> [31, 32, 33],
> [41, 42, 43],
> [51, 52, 53]]])
In this hypothetical case I'm selecting 3 windows of 3x3 within the main array (arr).
My actual array is a raster and I basically need a bunch (on the thousands) of little 3x3 windows.
Any help or even a hint will be much appreciated.
I actually haven't found any practical solution yet...since many many hours
THX!

We can leverage np.lib.stride_tricks.as_strided based scikit-image's view_as_windows to get sliding windows. More info on use of as_strided based view_as_windows.
from skimage.util.shape import view_as_windows
def select_random_windows(arr, number_of_windows, window_size):
# Get sliding windows
w = view_as_windows(arr,window_size)
# Store shape info
m,n = w.shape[:2]
# Get random row, col indices for indexing into windows array
lidx = np.random.choice(m*n,number_of_windows,replace=False)
r,c = np.unravel_index(lidx,(m,n))
# If duplicate windows are allowed, use replace=True or np.random.randint
# Finally index into windows and return output
return w[r,c]
Sample run -
In [209]: arr
Out[209]:
array([[ 0, 1, 2, 3, 4, 5, 6, 7, 8, 9],
[10, 11, 12, 13, 14, 15, 16, 17, 18, 19],
[20, 21, 22, 23, 24, 25, 26, 27, 28, 29],
[30, 31, 32, 33, 34, 35, 36, 37, 38, 39],
[40, 41, 42, 43, 44, 45, 46, 47, 48, 49],
[50, 51, 52, 53, 54, 55, 56, 57, 58, 59]])
In [210]: np.random.seed(0)
In [211]: select_random_windows(arr, number_of_windows=3, window_size=(2,4))
Out[211]:
array([[[41, 42, 43, 44],
[51, 52, 53, 54]],
[[26, 27, 28, 29],
[36, 37, 38, 39]],
[[22, 23, 24, 25],
[32, 33, 34, 35]]])

You can try [numpy.random.choice()][1]. It takes a 1D or an ndarray and creates a single element or an ndarray by sampling the elements from the given ndarray. You also have an option of providing the size of the array you want as the output.

Related

randomly sample from a high dimensional array along with a specific dimension

There has a 3-dimensional array x of shape (2000,60,5). If we think it represents a video, the 2000 can represent 2000 frames. I would like to randomly sample it along with the first dimension, i.e., get a set of frame samples. For instance, how to get an array of (500,60,5) which is randomly sampled from x along with the first dimension?
You can pass x as the first argument of the choice method. If you don't want repeated frames in your sample, use replace=False.
For example,
In [10]: x = np.arange(72).reshape(9, 2, 4) # Small array for the demo.
In [11]: x
Out[11]:
array([[[ 0, 1, 2, 3],
[ 4, 5, 6, 7]],
[[ 8, 9, 10, 11],
[12, 13, 14, 15]],
[[16, 17, 18, 19],
[20, 21, 22, 23]],
[[24, 25, 26, 27],
[28, 29, 30, 31]],
[[32, 33, 34, 35],
[36, 37, 38, 39]],
[[40, 41, 42, 43],
[44, 45, 46, 47]],
[[48, 49, 50, 51],
[52, 53, 54, 55]],
[[56, 57, 58, 59],
[60, 61, 62, 63]],
[[64, 65, 66, 67],
[68, 69, 70, 71]]])
Sample "frames" from x with the choice method of NumPy random generator instance.
In [12]: rng = np.random.default_rng()
In [13]: rng.choice(x, size=3)
Out[13]:
array([[[40, 41, 42, 43],
[44, 45, 46, 47]],
[[40, 41, 42, 43],
[44, 45, 46, 47]],
[[16, 17, 18, 19],
[20, 21, 22, 23]]])
In [14]: rng.choice(x, size=3, replace=False)
Out[14]:
array([[[ 8, 9, 10, 11],
[12, 13, 14, 15]],
[[32, 33, 34, 35],
[36, 37, 38, 39]],
[[ 0, 1, 2, 3],
[ 4, 5, 6, 7]]])
Note that the frames will be in random order; if you want to preserve the order, you could use choice to generate an array of indices, then use the sorted indices to pull the frames out of x.

Plot Network statistics using matplotlib

I try to use matplotlib to print network statistics. I want to look it like line graphs created with excel.
Excel:
Matplotlib
[
My very simple code:
import matplotlib.pyplot as plt
import numpy as np
x = np.array([0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59])
y = np.array(['0.00', '0.00', '0.00', '0.12', '0.00', '0.00', '0.00', '14.75', '108.56', '78.91', '508.15', '79.66', '147.84', '199.87', '14.02', '10.05', '3411.12', '19735.23', '19929.51', '18428.82', '21727.14', '19716.41', '20295.20', '20283.08', '20088.10', '20155.81', '20108.67', '19954.45', '20316.46', '20045.77', '20233.71', '19981.40', '20230.02', '20099.69', '20000.23', '20234.06', '19763.92', '20458.40', '19626.22', '20542.25', '19821.72', '20443.78', '20109.41', '19918.96', '20223.37', '19933.64', '20023.73', '19655.67', '19890.94', '20590.04', '20158.37', '20001.59', '20011.48', '19785.95', '20550.63', '19687.02', '20025.00', '20478.25', '20124.66', '20148.08'])
plt.plot(x, y)
plt.xticks(x)
plt.show()
Your y is string type. Try y=y.astype(float) before plot, then you get the expected:

Is there any possibility to Merage element in sublist with another one by using certain condition?

merge element of sublist with another sublist without duplicate
I am working on solving the vehicle routing problem and get the initial best solution and I want to be exploited and need to get rid of duplicate nodes while merging them.
my problem I want to merge the sublist of element [8] with others that carry the condition equal to 3 elements, but cause there are two sets consist 3 elements like [18, 22, 34, 8], [35, 36, 37, 8] and I need integrate element 8 in one of them randomly
bestsolution= [[22, 15, 20, 2, 32, 30, 4, 17], [27, 8, 9, 14, 33, 21, 5, 13], [26, 28, 6, 31, 11], [18,22,34],[35,36,37],[8]]
for a in bestsolution:
if len(a)==1:
p=a
del bestsolution[-1]
for b in bestsolution:
if len(b)==2:
b.extend(p)
print("p",b)
print("bestsolution1-2",bestsolution)
elif len(b)==3:
b.extend(p)
print("p",b)
print("bestsolution1-3",bestsolution)
my results:
p [18, 22, 34, 8]
bestsolution1-3 [[22, 15, 20, 2, 32, 30, 4, 17], [27, 8, 9, 14, 33, 21, 5, 13], [26, 28, 6, 31, 11], [18, 22, 34, 8], [35, 36, 37]]
p [35, 36, 37, 8]
bestsolution1-3 [[22, 15, 20, 2, 32, 30, 4, 17], [27, 8, 9, 14, 33, 21, 5, 13], [26, 28, 6, 31, 11], [18, 22, 34, 8], [35, 36, 37, 8]]
where for each run program, I got two results at the same time so how could be refuse the second solution.
'''
code targeted:
bestsolution1-3 [[22, 15, 20, 2, 32, 30, 4, 17], [27, 8, 9, 14, 33, 21, 5, 13], [26, 28, 6, 31, 11], [18, 22, 34], [35, 36, 37, 8]]
'''
Thank you
Ans/
The code will be:
'''
bestsolution= [[22, 15, 20, 2, 32, 30, 4, 17],[2,3,4], [27, 8, 9, 14, 33, 21, 5, 13], [26, 28, 6, 31, 11], [18,22,34],[35,36,37],[8]]
if len(a)==1:
p=a
del bestsolution[-1]
for b in bestsolution:
if len(b)==2:
b.extend(p)
print("p",b)
print("bestsolution1-2",bestsolution)
break
elif len(b)==3:
b.extend(p)
print("p",b)
print("bestsolution1-3",bestsolution)
break
elif len(b)==4:
b.extend(p)
print("p",b)
print("bestsolution1-4",bestsolution)
break
'''
Output expected:
p [18, 22, 34, 8]
bestsolution1-3 [[22, 15, 20, 2, 32, 30, 4, 17], [27, 8, 9, 14, 33, 21, 5, 13], [26, 28, 6, 31, 11], [18, 22, 34, 8], [35, 36, 37]]
'''
After trying much time I got final results, really didn't expect that I solved by putting the keyword (Break) will prevent execute the program in the second iteration so we will get one solution only.
if anyone have a comment, it will be a pleasure.

How to sort YOLOv4 bounding box?

I have trained the yolov4 from alexeyab darknet repo to detect characters in a number plate. It segments the character correctly but the bounding boxes are in random order. How can I sort the bounding box from top left to bottom right for image like this: (This is not actual image used but this is a photoshoped image for sample Nepali License Number Plate because of confidential data)
I've tried: (from pyimagesearch)
def sort_bbox(bbox, method="left-to-right"):
# initialize the reverse flag and sort index
reverse = False
i = 0
# handle if we need to sort in reverse
if method == "right-to-left" or method == "bottom-to-top":
reverse = True
# handle if we are sorting against the y-coordinate rather than
# the x-coordinate of the bounding box
if method == "top-to-bottom" or method == "bottom-to-top":
i = 1
# construct the list of bounding boxes and sort them from top to
# bottom
boundingBoxes = sorted(bbox, key=lambda b: b[1], reverse=reverse)
# return the list of sorted contours and bounding boxes
return boundingBoxes
but didn't sort the bounding boxes. It's still in random order.
I have bounding box from yolov4 detection like this: unsorted bounding boxes in xywh: [[50, 12, 15, 18], [66, 10, 15, 19], [87, 10, 19, 20], [21, 12, 24, 19], [51, 12, 15, 17], [51, 12, 15, 18], [66, 12, 15, 18], [86, 11, 19, 19], [39, 32, 27, 29], [68, 33, 28, 27], [97, 31, 28, 30], [12, 37, 24, 25], [11, 35, 25, 27], [40, 34, 27, 28], [68, 33, 27, 27], [97, 33, 28, 28]]
and from above sorting code: [[66, 10, 15, 19], [87, 10, 19, 20], [86, 11, 19, 19], [50, 12, 15, 18], [21, 12, 24, 19], [51, 12, 15, 17], [51, 12, 15, 18], [66, 12, 15, 18], [97, 31, 28, 30], [39, 32, 27, 29], [68, 33, 28, 27], [68, 33, 27, 27], [97, 33, 28, 28], [40, 34, 27, 28], [11, 35, 25, 27], [12, 37, 24, 25]]
What I want is the bounding box of: बा २ प ८ ८ ८ ८
Any Help will be very much appreciated.

How to update values in list of dictionaries

I want to update the value of a key in dictionary. This is a snippet of a list that contains over 300 dictionaries
chats = [
{'hour': 10, 'operator': 'john_doe', 'duration': [22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59], 'date': '2019-09-09'},
{'hour': 10, 'operator': 'john_doe', 'duration': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], 'date': '2019-09-09'},
{'hour': 10, 'operator': 'john_doe', 'duration': [18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28], 'date': '2019-09-09'},
{'hour': 11, 'operator': 'john_doe', 'duration': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28], 'date': '2019-09-09'},
{'hour': 10, 'operator': 'joseph_doe', 'duration': [5, 6, 7, 8, 9], 'date': '2019-09-09'}
]
script: I am getting an error on that script. I am looping to know if this dict is already in so that I can update the duration.
chat_list = list()
for chat in chats:
hour = chat.get('hour')
operator = chat.get("operator")
if len(chat_list) == 0:
chat_list.append(chat)
else:
found = False
for i in chat_list:
hour2 = chat.get('hour')
operator2 = chat.get("operator")
if (hour2 == hour) and (operator == operator2):
found = True
#concat both dictionary
i['duration'] = i.get('duration') + chat.get("duration")
if found == True:
found = False
else:
chat_list.append(chat)
My expected output is
chat_list = [
{'hour': 10, 'operator': 'john_doe', 'duration': [22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28], 'date': '2019-09-09'},
{'hour': 11, 'operator': 'john_doe', 'duration': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28], 'date': '2019-09-09'},
{'hour': 10, 'operator': 'joseph_doe', 'duration': [5, 6, 7, 8, 9], 'date': '2019-09-09'}
]
or
df = pd.DataFrame(chat_list)
df['duration'] = df['duration'].apply(lambda x: list(set(x)))
To be honest, I didn't tested your algorithm. Instead I took it as a small challenge and I wrote the following algorithm which doesn't need to copy chats in to a new list.
It finds the first occurrence of "similar" chat and concat the duration arrays. Then it deletes the "duplicated" chat. Further explanation in the code itself:
chats = [
{'hour': 10, 'operator': 'john_doe', 'duration': [22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59], 'date': '2019-09-09'},
{'hour': 10, 'operator': 'john_doe', 'duration': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], 'date': '2019-09-09'},
{'hour': 10, 'operator': 'john_doe', 'duration': [18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28], 'date': '2019-09-09'},
{'hour': 11, 'operator': 'john_doe', 'duration': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28], 'date': '2019-09-09'},
{'hour': 10, 'operator': 'joseph_doe', 'duration': [5, 6, 7, 8, 9], 'date': '2019-09-09'}
]
index = 0
while index < len(chats) - 1:
chat = chats[index]
# detect if there is another "similar" chat in the list (before this one)
first_index = next(
i for i, first_chat in enumerate(chats)
if chat.get('hour') == first_chat.get('hour') and chat.get('operator') == first_chat.get('operator')
)
# if the first index found is not this one:
# - concat `duration` arrays
# - delete this (duplicated) chat
if index != first_index:
chats[first_index]['duration'] += chat['duration']
del chats[index]
# otherwise continue and increment the index
else:
index += 1
print(chats)

Resources